Spatial analysis at the municipality and census unit levels present different spatial patterns. Figure 28 presents the cluster and outlier analysis for the municipality level and for the census unit level . Such analyses present multiple census units as outliers of low value in the center and northern areas of the map. Southern areas present clusters of lower value with areas of concern at the census unit level. Also, the end of the Altiplano is visible with the cluster of high values at the census unit level. The municipality level has a greater proportion of non-significant results and a cluster of high values in the northern part of the map. Another finding of interest is the similarity of the hotspot analysis at the municipality level with the impacts revealed in my first study . The actual temperatures of the region do not match the reported emergencies related to cold temperature . However, some spatial association is present between those impacts related to temperature that are not supposed to be extreme with hot spots of high social vulnerability. This raises the possibility that places with clusters of high social vulnerability might not require extreme weather conditions for impacts to occur. The level of analysis optimal for decision-making might vary according to who the information is supposed to aid and for what purpose. Therefore, one cannot identify a single “correct” scale of analysis. Furthermore, the results of this study suggest the need for a more detailed look at impacts resulting from weather events of lesser intensity and duration. Social vulnerability might be strongly spatially associated with such impacts. Moreover,grow table the reason that socio-economic indicators follow some patterns linked to topography is unknown; this pattern should be studied further.As seen in the previous chapter, spatial patterns of social vulnerability change somewhat with level of aggregation.

Due to the complexity in the data and the modeling of individual behaviors, spatially aggregated units are necessary; however, such units are typically not meaningful regions but administrative divisions . An administrative boundary is an echo of governmental relations , an abstract entity that defines the limits of administrative responsibility . These human-made borders are physically intangible in most cases but have tangible effects . However, they rarely correspond to variations in vulnerability . In geographic research, spatial units are usually determined by data availability rather than by the dynamics of one’s system of interest . This standard approach to vulnerability mapping is highly problematic insofar as we do not know the exact spatial configuration and boundaries of the geographical area that encompasses the phenomenon under study . Even though the goal of a sophisticated systems approach is to model at the individual level, sometimes it is optimal to aggregate data to units that offer a higher level of abstraction . Research conclusions and administrative decisions based on information that does not consider the effects of spatial scales and geographical boundaries may be misled and can have undesirable impacts .An approach that is independent of administrative boundaries was developed by Lang et al. . This relatively new approach to aggregation and regionalization is called the geon approach  Geons, in the context of social vulnerability assessment, are similar to the “geo-atoms” proposed by Goodchild et al. : a “representational primitive [that] has the capability to serve as the foundation for both object-based views through aggregation as well as field based views through a convolution operation using some discretization” . Geons are not the smallest or most atomic unit, as is the case with topons or chronons discussed by Couclelis . As such, geons do not claim to be undividable . A geon is designed to be as small as needed so it is maximally applicable in a given context .

Nor should geons be confused with the Elementary_geoParticle concept proposed by Voudouris . Geons are basic building blocks with no predefined substructure; Elementary_geoParticles possess the fixed definition of a cell . Geons are homogenous spatial objects generated by a scale-specific spatial regionalization of a complex and multidimensional geographical reality . Two types of geons appear in the literature: composite geons and integrated geons . Due to the abstractness of social vulnerability assessments, I use integrated geons. I aim to create homogeneous and integrative spatial regions independent from administrative units . The results present a variation of homogeneity that in some cases completely disregarded the municipality boundaries present in the region. In other instances, relation to those municipality boundaries is present. Figure 31 presents a set of maps using integrated geons for each capital as well as for the overall social vulnerability index. Mapping indices based on geons significantly fewer units than the original census units. The integrated geon methodology created at least 80% fewer zones than the census units. In the overall social vulnerability map, the methodology combines the 3,009 original census units into 500 geons. Homogeneity differs by domain—the maximum is 591 geons of financial capital, and the minimum is 331 geons of natural capital. As expected, neighboring farmers possess more homogenous levels of natural capital than of other capitals. A higher heterogeneity in financial capital is to be expected, given the diversities of income among individuals. In the next section, I discuss in detail each map and overlay municipality boundaries on them in order to examine heterogeneities inside municipality divisions. Figure 32 presents the lack of human capital for Puno using integrated geons, with and without municipality borders. The lack of human capital is high on the northwestern side of the region.

This is exacerbated in areas closer to the end of the Altiplano, where elevation begins to fluctuate. In contrast, northeastern and southeastern areas present better levels of human capital. Municipalities with areas that have very low levels of human capital tend to be low throughout and could be of concern to relevant authorities. Few municipalities present extreme values of human capital. Figure 33 presents the lack of social capital in Puno, with and without municipality borders. Most of the department of Puno has poor social capital with very few areas displaying better social capital levels. Compared to other capitals, social capital presents the most homogeneity in values across the entire department, and the values are fairly poor. An area of better social capital in the center west of the map is part of two neighboring municipalities. An interesting situation was the boundary between the extremes coincides with their municipality boundary. Figure 34 presents Puno’s lack of natural capital, with and without municipality borders. The lack of natural capital is strongest surrounding Lake Titicaca, mainly because numerous farms are less than five hectares. Only the municipalities surrounding Lake Titicaca have 100% of their territories with the lowest level of natural capital. The western outskirts of the region present middle levels of natural capital because the land there available for crops is limited. There is also low natural capital at the end of the Altiplano but only on the eastern side. It is intriguing to consider why natural capital shows this pattern,4×8 grow table with wheels but we cannot tell from the maps presented here. Figure 35 presents the lack of physical capital in Puno, with and without municipality borders. The lowest levels of this capital can be seen in the center and western areas of Puno, a pattern of great importance for decision-makers. In the southwestern outskirts, two neighboring municipalities have contrasting extreme values, the same pattern we observed with social capital. What explains these extremes in both physical and social capital across the municipality boundaries? Figure 36 presents Puno’s lack of financial capital, with and without municipality borders. Low levels of the financial capital are present in southeastern areas near the Bolivian border as well as northern areas toward Cuzco. Certain locations display better levels of financial capital, an important fact in a region dominated by poverty. Overlaying municipality boundaries on this map allows us to see which municipalities possess heterogeneous vs. homogenous levels of vulnerability. Most of the municipalities are heterogeneous inside their boundaries; however, municipalities with the worst values tend to Finally, Figure 37 presents the multivariate index of Puno’s overall social vulnerability, with and without municipality borders. The worst scores concentrate in the north with better values in the northeast and south. Overlaying municipalities reveals variation within municipalities, but extreme contrasts are not evident inside them. Geons often do not follow municipality divisions, but patterns following those boundaries was present in the maps. For example, both physical and social capital presented two neighboring geons displaying extremes of the index values. This study cannot cover the specific reasons or how significant is that certain vulnerabilities follow municipality boundaries. However, the delineation of the boundaries could be an indication that local policies and resources are either hindering or aiding these areas.

The population on the dataset are farmers and raises questions related to the entire population: are this difference only with farmers? Or the composition of the population affects the resources since one could be entirely farmers and the other a minority in the community? A study on why certain vulnerability patterns follow administrative regions would be imperative to implement similar policies on their neighboring areas. Furthermore, could provide information on what characteristics are causing the heterogeneity. Perhaps municipalities can work with their neighbors to address similar concerns. Furthermore, regional officials can use these maps to address concerns at the level of the phenomena and not just according to imposed administrative boundaries. However, the methodology is not easy to replicate due to its cost and required software. The methodology uses a software that its licensing could be costly to many universities and organizations. Future efforts should develop affordable ways for regional and national agencies in developing countries to apply this approach.Assessing all potential aspects of vulnerability is difficult. However, the literature is increasingly recognizing the role that social inequality plays in creating vulnerabilities. Human vulnerabilities must be a fundamental concern in developing and accessing disaster related policies. Therefore, a better understanding of the components that lead to vulnerability—understood within its local context—is imperative to reducing the negative consequences of hazard events, including those related to climate. This dissertation explored the importance of local context in shaping the indicators selected for vulnerability indices and the quantitative patterns of its spatial distribution. Using a mix-method approach combines the strength of both qualitative and quantitative methodologies while lessening its weaknesses. The popular purpose of triangulation, corroboration of the results with different methodologies, is not the only purpose for conducting mix-methodologies. This dissertation used a qualitative methodology to inform quantitative methodologies . The qualitative portion of this dissertation takes an inductive approach that it allows discoveries to arise unexpectedly due to the slow accumulation of evidence and provides entry points for intervention. Humans do not follow rational behaviors and understanding the local context allow us to observe where are those entry points. Furthermore, adding context to an asset-based assessment aids in understanding the information behind an indicator thus making interpretation meaningful. Ethnicity and native language reveal a complicated relationship to social vulnerability in this region, specifically among a population made up of farmers, 92% who identify as Quechuas or Aymaras. My results raise questions about using physical distance to health facilities, schools, and downtown areas as indicators of vulnerability. Education attainment is essential, and whether a farmer completes high school or not significantly influences levels of social vulnerability for his or her future. Years in school had different meanings for many farmers as a function of differences in the quality of education, differences that are understood to affect what knowledge is acquired in school. Another important component of social vulnerability and local context are the weather-related events and impacts a farmer has experienced. Every year farmers in the Peruvian Altiplano suffer and die due to cold temperatures in the region. Focusing only on heatwaves because of their increasing frequency and intensity around the world is analogous to ignoring farmers in rural locations because the world is urbanizing. Moreover, temperature-related events not only result in death but negatively impact livelihoods, health, and comfort. People should be able to do more than merely survive cold temperatures. Furthermore, stakeholders should pay attention to improving indoor thermal comfort, not just outdoor.